/sci/ - Science & Math » Thread #11174830

38KiB, 447x223, Anaconda_Logo.png

View Same Google iqdb SauceNAO

/sci/entific computing thread

Anonymous Mon 25 Nov 03:00:12 2019 No.11174830 View Reply Original Report

Quoted By: >>11174911 >>11174991 >>11175099 >>11175698

what kind of scientific tools do you use? data analysis and control software included.

inb4 labview, shit is for noobs

here are some i use:
>numpy
>scipy
>pandas
>numba
>matplotlib
>altair

Anonymous

Anonymous Mon 25 Nov 2019 03:02:15 No.11174833 Report

Quoted By: >>11174834

ok but everyone uses numpy

other than that i exclusively operate using jupyter and vs code

Anonymous

Anonymous Mon 25 Nov 2019 03:02:44 No.11174834 Report

Quoted By: >>11174903

>>11174833
so you never make plots then?

Anonymous

Anonymous Mon 25 Nov 2019 03:25:27 No.11174874 Report

Quoted By: >>11174898

In order of preference

> Mathematica
> Julia
> Specialty library in python/c++ someone else has already written

Anonymous

Anonymous Mon 25 Nov 2019 03:39:28 No.11174898 Report

Quoted By:

>>11174874
>Mathematica
yes, Mathematica is clutch. but i find the only people who use it as their “primary” coding “language” are usually very computer-illiterate... like theoretical physicists who take a year to catch up once they leave science for industry

Anonymous

Anonymous Mon 25 Nov 2019 03:41:50 No.11174903 Report

Quoted By:

>>11174834
Yes.

Anonymous

Anonymous Mon 25 Nov 2019 03:45:16 No.11174911 Report

Quoted By: >>11174936

>>11174830
I use Nastran and occasionally abaqus. I click a lot of buttons. Don't ask me how to program DMAPs or subroutines.

Anonymous

Anonymous Mon 25 Nov 2019 03:55:26 No.11174936 Report

Quoted By: >>11174969

>>11174911
>Nastran and occasionally abaqus
i take it you work for the space administration agency of a major north american country. after looking up those two software packages, i am highly surprised you guys are still using such old software.

then again i heard that the US's nuclear launch systems were only migrated from relying on floppy disks in like the last few weeks.... so i am not that surprised that modern governmental space agencies are stuck in a 30-years-ago time-warp

Anonymous

Anonymous Mon 25 Nov 2019 04:06:33 No.11174969 Report

Quoted By: >>11175028

>>11174936
I'm not anywhere near anything that sounds as glamorous as you have put it. Even if I were in one of those magical places like NASA JPL, I strongly doubt it's as wonderful as pop culture portrays it. It hurts me a little bit to read those words since they remind me of a lot of hopes and promises that fell short.

>i am highly surprised you guys are still using such old software.
There is still an intense effort to develop Nastran and Abaqus. It's been long bought up by corporations (well mostly for Nastran), and they're trying to implement more and more into their solvers in order to tout as many features as they can.

Anonymous

Anonymous Mon 25 Nov 2019 04:13:30 No.11174991 Report

Quoted By: >>11175028

>>11174830
>Matlab with simulink
>Octave

Matlab is obviously king, octave has some very cool flexibility when writing scripts but can be very annoying at times and its very badly supported.

Anonymous

Anonymous Mon 25 Nov 2019 04:35:25 No.11175028 Report

Quoted By: >>11175063

>>11174969
sure, corporations are known for keeping dying software alive. is it still true that COBOL coders are the highest paid on average?

>>11174991
sorry anon, MATLAB is also for noobs. hate to break it to you. you can do everything MATLAB does with numpy and scipy and just as fast if you use numba

Anonymous

Anonymous Mon 25 Nov 2019 04:42:36 No.11175033 Report

Quoted By: >>11175041

Why is this on /sci/?

Anonymous

Anonymous Mon 25 Nov 2019 04:47:45 No.11175041 Report

Quoted By:

>>11175033
>why is scientific computing on the science board?
undergrad who does everything with pencil and paper detected

Anonymous

Anonymous Mon 25 Nov 2019 05:01:11 No.11175063 Report

Quoted By: >>11175072

>>11175028
>>sure, corporations are known for keeping dying software alive.

These aren't dying codes. Far from it. Still very actively developed, marketed, and supported. I don't know any other solvers that have come up recently that can do as much as the older and established solvers like Nastran, Abaqus, or Ansys.

Anonymous

Anonymous Mon 25 Nov 2019 05:08:25 No.11175070 Report

Quoted By:

$<br />\text{ }^{\color{#571da2}{\displaystyle\text{W}}}\text{ }^{^{^{^{\color{#462eb9}{\displaystyle\text{h}}}}}}\text{ }^{^{^{^{^{^{^{\color{#3f47c8}{\displaystyle\text{y}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{\color{#3f62cf}{\displaystyle\text{ }}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#437ccc}{\displaystyle\text{i}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#4b90bf}{\displaystyle\text{s}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#56a0ae}{\displaystyle\text{ }}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#62ab99}{\displaystyle\text{t}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#71b484}{\displaystyle\text{h}}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#82ba70}{\displaystyle\text{i}}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#96bc5f}{\displaystyle\text{s}}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#a9bd52}{\displaystyle\text{ }}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#bcbb48}{\displaystyle\text{o}}}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#ceb541}{\displaystyle\text{n}}}}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#dcab3c}{\displaystyle\text{ }}}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{^{^{^{\color{#e39938}{\displaystyle\text{/}}}}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{^{^{^{\color{#e68033}{\displaystyle\text{s}}}}}}}}}}}}\text{ }^{^{^{^{^{^{^{\color{#e3632d}{\displaystyle\text{c}}}}}}}}}\text{ }^{^{^{^{\color{#de4227}{\displaystyle\text{i}}}}}}\text{ }^{\color{#da2121}{\displaystyle\text{/}}}<br />$

Anonymous

Anonymous Mon 25 Nov 2019 05:09:14 No.11175072 Report

Quoted By: >>11175084 >>11175101

>>11175063
dude, it's written in FORTRAN. come on.

look, NASA collaborates with the particle physics and medical physics communities on the GEANT software package. and GEANT ported itself to C++ once they realized FORTRAN is a thing of the past. nothing written in FORTRAN can legitimately be considered modern in 2019

Anonymous

Anonymous Mon 25 Nov 2019 05:13:43 No.11175084 Report

Quoted By:

>>11175072
(by "it" i was referring to Nastran. i didn't check Abaqus or Ansys but the same applies if they are written in FORTRAN)

Anonymous

Anonymous Mon 25 Nov 2019 05:34:05 No.11175099 Report

Quoted By:

>>11174830
Algorithms and operational code is in Fortran, result analysis is in Python (the usual libraries), I occasionally do the small odd job with Julia, just do mix things up and learn new stuff. Then bash to start things and push jobs onto clusters.

Something like >90% of my working time is done on Emacs and ssh terminals, the rest is looking at plots and banging my head aganst nearby walls. Paper writing is TeXshop and Overleaf.

Anonymous

Anonymous Mon 25 Nov 2019 05:36:00 No.11175101 Report

Quoted By: >>11175117

>>11175072
Yes, I know it's in fortran, but finite element methods is still an area of research and development and it's not simply reducible to a matter coding language to gain speed or improve usability.

Alright consider this. What other alternative finite element software is Dassault Systemes working on other than Abaqus?

>look, NASA collaborates with the particle physics and medical physics communities on the GEANT software package. and GEANT ported itself to C++ once they realized FORTRAN is a thing of the past. nothing written in FORTRAN can legitimately be considered modern in 2019
This maybe true for particle physics and medical applications, but structural analysis is also a huge area where finite element analysis is employed.

Anonymous

Anonymous Mon 25 Nov 2019 05:52:14 No.11175117 Report

Quoted By: >>11175133

>>11175101
>it's not simply reducible to a matter coding language
i don’t see how this is an argument. if the language doesn’t matter, then why not port it to a modern language that has a broad community?

if you have legitimate arguments for why it only works in FORTRAN then that would be an argument, except i doubt it would be defensible given the wide array of modern programming languages that all have their individual strengths.

my guess is that Dassaulr and whoever else uses these kinds of software keep using them solely because of culture and/or a small community of guys (who may be very good at one thing or another but) who are stuck in the past. hanging onto FORTRAN is a recipe for death in the software world

Anonymous

Anonymous Mon 25 Nov 2019 06:02:51 No.11175133 Report

Quoted By: >>11175144

>>11175117
>why not port it to a modern language that has a broad community?
Within the scientific community, and especially the section that requires fast code, Fortran has a very broad community. Finding a scientist that can read and write C++ code is somewhat rare. Fortran is also very legible, doesn't (really) have pointer arithmetic, and supports native ND array operations and slicing, again without the use of pointers. I recently had to write a new piece of code from scratch, and ended up using Fortran, simply because that's what most people in my field know, and it doesn't require a CS dude to work through layers and layers of abstractions and templates if something breaks.

Anonymous

Anonymous Mon 25 Nov 2019 06:12:48 No.11175144 Report

Quoted By: >>11175157

>>11175133
>Within the scientific community, and especially the section that requires fast code, Fortran has a very broad community. Finding a scientist that can read and write C++ code is somewhat rare.
i don’t know what branch of science you are in but in mine, C++ is way more common than FORTRAN. and it is more common outside of science too. i suspect you are a generation or two older than the average /sci/ anon (also the average scientist, remembering that tenured boomers are that very small triangular piece on top of many layers of younger people)

Anonymous

Anonymous Mon 25 Nov 2019 06:21:19 No.11175157 Report

Quoted By:

>>11175144
>i don’t know what branch of science you are in but in mine, C++ is way more common than FORTRAN.
Earth systems, remote sensing and biosphere modelling. There are C++ projects out there, but they're the minority. Fortran (not uppercase since the 90s) still beats C++ for legibility, and unless runtime-polymorphism is critical to your application, it'll do just fine. I'm not quite an oldfart yet .. closer to 30 than to 40. Fortran might die at some point, who knows. For now, I'd still rather write something like "my_array[5:45] = result_x[0:40] + result_y[10:50]" than whatever the C++ equivalent is.

Anonymous

Anonymous Mon 25 Nov 2019 06:25:23 No.11175163 Report

Quoted By: >>11175171

Fortran (FTW)
C/C++
Bash
Gnuplot
Python

To a lesser extent:
Mathematica
IDL (Astronmy)

Anonymous

Anonymous Mon 25 Nov 2019 06:28:51 No.11175171 Report

Quoted By:

>>11175163
*Astronomy

Anonymous

Anonymous Mon 25 Nov 2019 06:30:47 No.11175173 Report

Quoted By:

Fortran (FTW)
C/C++
Bash
Gnuplot
Python

To a lesser extent:
Mathematica
IDL (Astronomy)

Anonymous

Anonymous Mon 25 Nov 2019 13:30:31 No.11175698 Report

Quoted By: >>11175707

>>11174830
Why the hell would anyone use Anaconda? Just install the packages without some bloated shit.

Anonymous

Anonymous Mon 25 Nov 2019 13:32:39 No.11175707 Report

Quoted By:

>>11175698
i agree, conda is bloated. but anaconda (the organization) develops a lot of the good tools available. and you can install all of them with pip instead if you like

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches